meshQualityDict: minVol for snappyHexMesh
 

Hello,

I'm calculating the minVol in system/meshQualityDict with the following formula:

minVol = a^3 / 8^lvl / x

where
a = size of a blockMesh element
lvl = maximum refinement level
x = some arbitrary number

It seems that for values of x lower than 15-20 there are too many too small cells to produce a good mesh.

So my question is:
What should i choose for x, and why?

Another question is, why is the minimum volume so important? Can I make a mesh with a too high resolution? The intuitive thinking would be that more elements mean more precise discretisation hence more exact results. Or is my thinking wrong?

Thanks!

-- Douglas Adams

Quick answers:
The simplest reason is that it can be critical for transient simulations (time-dependent), due to the Courant Number. Re-read the first tutorial on the OpenFOAM User Guide, where this is briefly explained.

It's possible to have "too much of a good thing"...

... specifically in CFD, it's advisable to perform mesh convergence tests when ascertaining the accuracy. A mesh too refined can easily reveal the limitations of the models being used. A mesh refined in the wrong places can lead to problematic/non-physical results.

See this blog post for a few examples: OpenFOAM: Interesting cases of bad meshes and bad initial conditions

Also, look up (Google?) the difference between mesh requirements between DNS, LES and RANS (or RAS).


Unfortunately not enough people are available at all times here at the forum, but with luck a few people are available a few times per week/weekend ;)

I know you're not obliged to answer anything and I appreciate any kind of help, I just made an elaborated 'bump'. IMO it's a good quote :)

Thanks (for now), let me do some googling...